What "AI readability" measures here
Traditional readability formulas (like Flesch-Kincaid) estimate how easy text is for a human to read. This tool instead checks structural signals that make text easier for an AI system to parse and extract facts from when summarizing or citing a page: sentence length, paragraph length, the presence of lists and headings, and passive-voice usage. It is a heuristic estimate based on well-documented content structure best practices, not a connection to any specific AI model's internal behavior.
Why human readability and AI parseability aren't quite the same thing
A human reader can follow a long, winding sentence with the help of tone, inflection, and general world knowledge - but a system extracting discrete facts for a summary tends to do better with short, single-idea sentences it can isolate cleanly. The two goals overlap substantially but aren't identical, which is why this tool checks a distinct (though related) set of structural signals from a traditional human-readability formula.
Why passive voice specifically gets flagged
A sentence like "the cause was identified as X" buries who or what did the identifying, while "researchers identified X as the cause" states it directly - active voice generally makes the actor and the action explicit in a way that's easier to extract as a clean fact, which is why a noticeably high passive-voice ratio lowers this tool's score.
Why question-style lines count toward the score
Framing a section around the exact question a reader (or an AI system matching a query to your page) would ask - "What is X?", "How does X work?" - as its own line gives that section a clear, literal match target. This tool checks for at least one line written as a direct question, since that pattern is one of the more commonly cited structural signals in answer-engine optimization write-ups, alongside headings and lists. It is a cheap, honest proxy for "is this content organized the way an FAQ or Q&A page would be" - not a measure of whether your questions are well-chosen.
Why a perfect score is a structural signal, not a citation guarantee
Whether an AI system actually cites or surfaces a specific page depends on many factors entirely outside text structure - the site's authority, how relevant the content is to a specific query, how fresh it is, and more. This tool checks only whether your content's structure makes it easy to parse if it is considered, which is a necessary condition worth getting right, but not a sufficient one for being cited.